[arch-dev-public] WARNING: libjpeg rebuilds are entering [testing] - devs get building!

Eric Bélanger snowmaniscool at gmail.com
Wed Jul 8 18:32:30 EDT 2009


On Wed, Jul 8, 2009 at 3:23 AM, Eric Bélanger<snowmaniscool at gmail.com> wrote:
> On Wed, Jul 8, 2009 at 3:13 AM, Gerardo Exequiel
> Pozzi<vmlinuz386 at yahoo.com.ar> wrote:
>> Eric Bélanger wrote:
>>> On Wed, Jul 8, 2009 at 1:08 AM, Pierre Schmitz<pierre at archlinux.de> wrote:
>>>
>>>> On Wednesday 01 July 2009 14:12:05 Allan McRae wrote:
>>>>
>>>>> Devs, please remember to mark off rebuilt packages from the TODO list as
>>>>> you go.
>>>>>
>>>> sane-frontends and netpbm do not compile.
>>>>
>>>>
>>>
>>> I just commited a patch for netpbm  to fix some of the compile
>>> problems. It still needs more patching to compile. The remaining error
>>> is:
>>>
>>> jpeg.c: In function `ReadJPEG':
>>> jpeg.c:473: error: `struct jpeg_decompress_struct' has no member named
>>> `min_DCT_scaled_size'
>>>
>>> This is because min_DCT_scaled_size is no longer defined in libjpeg's
>>> header files. I don't know how to fix it.  I wasn't able to find any
>>> patches. It seems that the libjpeg update is too new.
>>>
>>>
>> Reading the libjpeg.h seems that replacing with min_DCT_v_scaled_size is
>> fine. Quote from header: "Therefore, the IDCT output contains
>> v_samp_factor*DCT_v_scaled_size sample rows of a component per iMCU row."
>>
>> buffer_height = cinfo.max_v_samp_factor * cinfo.min_DCT_v_scaled_size;
>>
>> But maybe I wrong.
>>
>>
>> BTW: should be compiled with -fno-strict-aliasing there are some
>> warnings about this, and the code can or not work correctly.
>>
>
> I'll check it out.  The pkgrel for netpbm is misused. It'll need to be
> fixed. If someone as a better/verified patch, post it here. I'll take
> care of the rebuild and pkgrel fixing.

A rebuilt netpbm with this fix will be in testing soon. I'll
check/rebuild the sane packages.


>
> BTW, I just pushed rebuilds for celestia and kismet (x86_64 only).
> I'll do them for  i686  tomorrow unless someone does it before.
>


More information about the arch-dev-public mailing list